using predictive analytics for strategic planning at ... · crawford&&&company&...
TRANSCRIPT
Using Predictive Analytics for Strategic Planning at Crawford GTS
Predic've Analy'cs World, San Francisco
April 16, 2013
Dr. Andries Willemse – SVP Crawford GTS, Crawford & Company Sergo Grigalashvili – VP ICT , Crawford & Company
Crawford & Company
What is Global Technical Services SM (GTS)?
v Crawford’s defini've offering to major incident claims handling, embracing catastrophic and complex claims, high value property and casualty claims as well as specific claims such as construc'on, power and engineering
v It is the single best global resource that serves reinsurers, interna'onal corporate market, major loss / catastrophe divisions of composite insurers such as the London market, Lloyd’s syndicates as well as cap'ves, risk managers and their brokers
v A unique combina'on of experienced and qualified professionals, infrastructure and leadership enables us to strategically manage losses anywhere on the globe
2
Crawford & Company
Resource Management Challenge
v More than 500 senior adjusters and industry specialists opera'ng out of 144 loca'ons worldwide
v Serve 12 industry sectors
v Demand dependent on catastrophic natural and man-‐made events, market condi'ons, and individual professional rela'ons
3
Crawford & Company
Focused Trends, Seasonality, Forecasts
4
v Focus on an industry sector or subsector – Do we have sufficient expertise in the industry sector to meet the future demand? Recruitment and retention practices, specific industry knowledge, retirement and resignations & cross discipline training
v Focus on a region or country – Are we under/overstaffed in the region/country to meet the future demand? Business as usual vs CAT type events
v Focus on an insurer – Client specific demands in terms of contractual arrangements? Service level agreement compliance
v Focus on a product or peril – Are our peril-specific processes sufficiently optimal to handle future demand?
v Focus on an adjuster – Sufficiently aware of high risks to the organization / opportunities to improve? Improve employee relations; be proactive rather than reactive
Crawford & Company
Approach
5
v Tools – Proprietary loss data
– Data exploration
– Historic trends
– Forecasts
– Dashboards
v An'cipated benefits – Strategic planning
– Capacity-demand analysis
– Improved client relations
Large Loss Data
Adjuster Data
Forecasting
Engine
Strategic planning Capacity / demand
Client analysis
Economic, Industry
Data
Crawford & Company
Massive Scale
6
v Three variables being forecast – Claim count, total reserves, total fees
v For each for the following categories – All claims, catastrophe claims, non catastrophe claims
v For each of the following series – Claims everywhere
– 6 regions
– 12 industries
– 8 product lines
– 40 perils
– 54 industry subsectors
– Top 50 lead insurers
– Top 50 countries
– Top 50 adjusters
– Monthly forecast for the next 18 months; yearly aggregate forecasts through 2015
Crawford & Company
Tools Used
7
v Crawford GTS® proprietary Global Large Loss Database
v Published economic and industry data
v Available and inexpensive tools
Large Loss Data
Adjuster Data
Economic, Industry Data
Data consolida'on, storage, management
Sta's'cal forecas'ng: ARIMA, Holt-‐Winters, ETS
Forecast model automa'on; extended predic've modeling (s'll work in progress)
Data visualiza'on and explora'on
Crawford & Company
Step 1: Understand Trends, Seasonality
8
Select any series Trend vs.
seasonal vs. random
Actual trend, linear regression, 95% confidence band
Seasonal averages, and seasonality trend
v Analysis done for Claim Count, Reserves and Fees – By region and country, product and peril, industry sector and subsector, insurer, adjuster
– Note: Crawford GTS® book of business, not industry in general
Crawford & Company
Step 2: Forecast
9
80% confidence intervals
Three forecasting techniques
Select any series
v Analysis done for Claim Count, Reserves and Fees – By region and country, product and peril, industry sector and subsector, insurer, adjuster
– Note: Crawford GTS® book of business, not industry in general
Crawford & Company
Where Will Growth Be ? Decline ?
10
v Analysis done for Claim Count, Reserves and Fees – By region and country, product and peril, industry sector and subsector, insurer, adjuster
– Note: Crawford GTS® book of business, not industry in general
Forecasted increase in flood claims
Forecasted decrease in public liability claims
Crawford & Company
Capacity Analysis
11
v Forward looking analysis of capacity to meet the demand – Where are the risky areas ? Adjusters close to retirement.
– Where is the adjuster’s book of business concentrated ?
– Analyze by product, insured, geography, industry
Roll Forward
Crawford & Company
Capacity & Demand Predic'on
12
v Forward looking analysis of capacity vs. demand – When and where we need to hire or train; specific risk areas
– By country and industry sector Supply vs.
demand trend contrast
Forward looking age profile
Crawford & Company
Before and A`er
13
Before – reactive After – proactive
Recruit when we see shortage Recruit in anticipation of shortage; customize training based on future demand
React to claims coming in Be prepared based on seasonal factors and overall trends by industry and geography
Analyze impact after a key resource leaves
Analyze risk by looking at resource age profile, geography, industry expertise, product line, and client relationship; mitigate the risk of the resource leaving
Analyze impact after a key client leaves
Analyze risk by looking at client specific profile and forecast; mitigate the risk of the client leaving
Plan and budget based on educated guess and past experience
Use forecast models in conjunction with human judgment
Crawford & Company
Next Steps
14
v Define aggregate metrics to further simplify consump'on of the forecasts – Avoid information overload; highlighting important areas
v Periodically publish predic've analysis to help regional leadership – Break down by country, industry sector and subsector, product and peril, insurer
v Automate forecast models for weekly near term reforecas'ng
v Expand scope to include value, benefits, and profitability analysis – In-house costs vs. external; fixed costs vs. variable; client value metrics; what-if analysis
v Enrich the models with macro-‐economic factors and natural disaster forecast
– GDP growth, premium per capita, insured asses value, industrial production, insurance penetration – Government forecast of geophysical, meteorological, hydrological, and climatological events
Crawford & Company
Lessons Learned
15
v Crawford GTS® Large Loss Database – Leverage the value of proprietary data
v Lessons learned – We all know what happened in the past, focus on the future even if not clear – Contrary to how it looks, this does not have to be expensive – Data visualization is extremely helpful – Focus on the right business objectives and manage change – There is more randomness in life than we like to think
Thank You !
Dr. Andries Willemse – [email protected] Sergo Grigalashvili – [email protected]